Modeling the spread of smoking in adolescent social networks

Objectives: Using a systems-level approach, develop a large-scale, individual-based, simulation framework for an-alyzing the spread and control of smoking behavior in adolescent populations. Methods: We create a synthetic information environment that combines Add Health data with synthetic population data to create a model of smoking behavior diffusion in middle and high schools. This also involves generating synthetic friendship networks based on real friendship networks from the Add Health data. Simulations are done for a three-year period, using an efficient parallel simulation engine. Results: We derive diffusion curves by smoking rate, age, gender, and parents ’ smoking status, separately for the middle and high schoolers. We also compare them with similar diffusion curves obtained from the real friendship networks from Add Health. Conclusions: This research builds a synthetic laboratory for social epidemiologists where a rich representation of individual characteristics and contextual factors can be taken into account to understand the global dynamics of a variety of individual behaviors pertaining to smoking.